2024 · 13 citations · 8 references
ComputingComputer ScienceComputer SystemsArtificial IntelligenceInternet Of ThingsMachine LearningDeep LearningFog ComputingReinforcement LearningTask OffloadingDeep Reinforcement LearningFog Computing ArchitectureFog Task OffloadingOn-policy LearningQ-learningQos-aware SchedulingReinforcement Learning For OffloadingService Offloading
The proliferation of smart devices has led to a rapid expansion of Internet of Things (IoT) applications. Consequently, a substantial volume of unprocessed data is produced, necessitating processing and storage. Individual IoT devices alone are insufficient to manage significant volumes of data. Fog computing is an essential technology that enables the proximity of cloud services to the end-user. Efficient task offloading techniques are necessary to balance the demand due to the unpredictable nature of tasks with user's Quality of Service (QoS) needs. To address the issue effectively, two different model-free off-policy i.e. Deep Q-Network (DQN) and on-policy i.e. Asynchronous Advantage Actor-Critic (A3C) Deep Reinforcement Learning (DRL) techniques are described as optimising resource utilisation by maximising the reward. Ultimately, extensive experiments are carried out to verify and assess the efficiency and effectiveness of the proposed methods.
8
A Survey and Taxonomy on Task Offloading for Edge-Cloud Computing
Bo Wang, Changhai Wang, Wanwei Huang et al. · IEEE Access · 2020 · 149 citations · Full text